In one embodiment, a graphics processor includes a register file having a plurality of storage segments to store information and output a plurality of segment outputs via a plurality of segmented bitlines to a static logic circuit to receive the plurality of segment outputs from the plurality of storage segments and to output read data based on the plurality of segment outputs. The register file may output the read data with a same amount of power without regard to a logic state of the read data. Other embodiments are described and claimed.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A processor comprising: a streaming multiprocessor comprising: a register file comprising a plurality of storage segments to store information and output a plurality of segment outputs via a plurality of segmented bitlines to a static logic circuit, the static logic circuit to output data based on the plurality of segment outputs, wherein the register file is to select a storage segment of the plurality of storage segments for storage of the data based at least in part on read frequency of the data, and store the data in the selected storage segment of the plurality of storage segments; a plurality of arithmetic logic units coupled to the register file; a plurality of texture units coupled to the register file; and a shared memory coupled to the plurality of arithmetic logic units and the plurality of texture units.
2. The processor of claim 1 , further comprising scheduler logic to schedule groups of instructions.
3. The processor of claim 1 , wherein the plurality of arithmetic logic units are to perform operations on integer data types.
4. The processor of claim 1 , further comprising at least one memory unit.
5. The processor of claim 4 , wherein the memory unit comprises a load and store unit.
6. The processor of claim 1 , further comprising at least one special function unit.
7. The processor of claim 1 , wherein the streaming multiprocessor comprises a single instruction multiple thread processor.
8. The processor of claim 1 , further comprising a graphics processing unit comprising the streaming multiprocessor, the graphics processing unit to couple to a host processor via a high speed interconnect.
9. The processor of claim 8 , wherein the host processor and the graphics processing unit are integrated in a single package.
10. The processor of claim 1 , wherein the plurality of storage segments comprises: a plurality of bitcells each to store a data bit; a first local bitline coupled to a first collection of the plurality of bitcells to output read data from the first collection of the plurality of bitcells; a second local bitline coupled to a second collection of the plurality of bitcells to output read data from the second collection of the plurality of bitcells; and a latch circuit having a first input coupled to the first local bitline and a second input coupled to the second local bitline, the latch circuit to output a segment output, the segment output comprising a static signal.
11. The processor of claim 1 , wherein the register file comprises a single-ported register file comprising hierarchical local and global read bitlines.
12. The processor of claim 1 , further comprising a mapping circuit to store first information associated with a first thread having a first priority in a first storage segment of the plurality of storage segments and store second information associated with a second thread having a second priority in a second storage segment of the plurality of storage segments, the first priority higher than the second priority.
13. The processor of claim 1 , further comprising a plurality of transmission interconnects to couple the plurality of storage segments to the static logic circuit, wherein at least some of the plurality of transmission interconnects have differing lengths.
14. The processor of claim 1 , wherein the register file is to consume a same amount of power to read a logic one value as to read a logic zero value.
15. A method comprising: receiving, in a control circuit associated with a memory, an incoming write request for data to be written to the memory, the incoming write request including a first address of the memory; re-mapping, by the control circuit, the first address to a second address of the memory, based at least in part on a priority of the data; and storing the data in a second segment of a plurality of segments of the memory, the second segment associated with the second address.
16. The method of claim 15 , wherein the second segment is associated with a lower power consumption than a first segment of the plurality of segments, the first segment associated with the first address.
17. The method of claim 16 , further comprising reading the data from the second segment, with reduced power consumption than reading second data from the first segment.
18. The method of claim 15 , wherein the priority of the data comprises quality of service information for a thread.
19. An integrated circuit comprising: one or more cores; and a register file coupled to the one or more cores, wherein the register file comprises a plurality of storage segments including a first storage segment and a second storage segment, wherein a read operation to the first storage segment is to consume less power than a read operation to the second storage segment.
20. The integrated circuit of claim 19 , wherein the register file is to store first data in the first storage segment based at least in part on read frequency of the first data.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
June 10, 2019
September 29, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.